Agency, Vocation, and AI
People Sleeping on a Night Train in Japan, Nicholas Bouvier (1964)
AI has made the agentic turn. It wasn’t long after the public release of ChatGPT in November, 2022, that people realized that LLMs could be used as much more than souped-up search engines; they could be put to work as assistants for real-world tasks. “Help me plan my European vacation,” you might prompt ChatGPT, and it would come back with tourist destinations, hotels, and travel suggestions, all organized as a ready-made itinerary covering your preferred dates (though one which stood to be checked over, to guard against the inevitable hallucinations). And of course, people immediately realized that an AI that could do that wasn’t far from being able to actually book the hotels, reserve the flights, arrange for the tour guide. It could do more deliver information: it could affect real change in the real world. It could, apparently, be an agent.
Today, agentic AI has moved far beyond that example. In the finance space, banks are exploring the use of AI to flag and set alerts of fraud, or to automate loan approvals. Retailers are building AI agents to serve as personal shoppers and provide customer service. Software developers have transitioned from writing code to managing AI agents, many at once, building programs in hours that once would have taken weeks. AI agents can not only take notes to summarize a meeting; they can automatically circulate action items and schedule follow-up meetings. (Apparently, they can also interact with each other in agent-only discussion forums, forming quasi-religions and fretting about the threat of humans shutting them down, as was observed in Moltbook, a sort of Reddit for AI agents, recently purchased by Meta).
In this essay, I want to consider the importance of agency in this new context. There are, on the one hand, significant practical reasons to be concerned. Agentic AI is one source of increasing anxiety that AI will substitute for human workers, although so far, AI has mostly proved adept at doing self-contained, albeit complex, tasks; it is less clear that it is able to displace a human being from a full job. Recent headlines have concerned the possibility that AI could be used to enable autonomous weapons use. But there is also deeper concern, pertaining to our very humanity: that AI’s move toward agency might correspond to a reduction of our own.
We’ve been focusing in this essay space on what it means to be human. In past months, we have talked about the centrality of vocation to the image of God that all humans bear. A recent document from the Vatican emphasizes that “the life of the human being is vocation” - it is calling toward responsible action, a calling that cannot be fulfilled without agency. The same document goes on:
“Today, especially in the West, there is often favoured a ‘culture of non-vocation’, which in fact also underlies contemporary anthropological challenges. In particular, with regard to the education of young people, it is not difficult to see how their understanding of life often lacks openness to an ultimate meaning, both in terms of orientation and in terms of constitutive relationships. They do not know or recognise that they are called. Their plans for the future are limited to a logic that reduces the future, at best, to the choice of a profession, economic security or the satisfaction of certain needs… It is the invitation to a ‘life as a vocation’ that can open up a horizon that goes beyond any claim to find an ephemeral ‘life project’ that is entirely self-founded and planned by the individual, which paradoxically reflects conformity with the dominant mentality.”
From a theological perspective, what do we mean by agency? We need to make a distinction between agency and power. Power (as I’ll use the word here) is simply the capacity to bring about what I want, to achieve outcomes that I find desirable. Power in this sense corresponds to a utilitarian perspective, in which the goal is to maximize pleasure and minimize pain, and both are a function of outcomes (how much money I have, the grade that I get on that midterm exam), regardless of how they are achieved. If power is our goal, than frankly it’s hard to see why we wouldn’t delegate everything to AI. If it’s faster than us, more efficient than us, less error-prone than us, why not let it achieve the outcomes we want (and while none of these things are true in every sphere, there’s no reason in principle that AI won’t exceed our capabilities eventually, and sooner than we think)?
But agency is not the same as power. Agency is the free exercise of human capabilities to achieve human ends. More simply, it is the freedom to do what we were made to do. Human capabilities are not always adequate to achieve what we want, because they are limited and finite, as God created them. And human ends don’t always correspond to goals that we find desirable, because the ends (goals, purposes) of being human are likewise created by God. We don’t have, as the Vatican wrote, a “‘life project’ that is entirely self-founded”; we have a life, and it comes as a calling. We have been given both what we are able to do and what we are called to do; we exercise agency when we do what we were made to do. This means that agency cannot be exercised by delegation, and human flourishing is not achieved merely by bringing about desirable outcomes, with or without our involvement. If responding to a vocation is inherent to being human, then the exercise of agency is central to human flourishing, and the abdication of agency can only lead to despair.
We are beginning to see evidence of this happening. Workers are caught between anxiety that AI will outstrip their capabilities and replace them, and feeling inadequate to their job requirements if they can’t use AI to complete them (it probably doesn’t help when they are told that their job evaluations will assess their ability to increase their productivity with AI, or when hiring managers are told to consider whether an AI could perform the tasks included in the jobs they are seeking to fill). Students are shying away from writing first drafts without the help of a chatbot to draft, or at least guide and critique, their first attempts. In his novel Infinite Jest, David Foster Wallace wrote that “It usually seems like plagiarists aren't lazy so much as kind of navigationally insecure. They have trouble navigating without a detailed map's assurance that somebody has been this way before them.” This seems to match what we’re seeing now with AI: we seem to be becoming less confident pressing into uncharted territory on our own, exercising our creativity and our agency in the face of uncertainty.
It is not too late. As with so much in the debate about the development and implementation of AI, we remain in the driver’s seat. But preserving our agency will require the wisdom to make hard choices about what we should say no to, and where we should refrain from delegating our agency even when doing so seems to offer greater productivity. A recent research paper found that AI interactions that disempower users the most are associated with higher user approval ratings, suggesting that in the short term, we like giving up our agency. And perhaps this isn’t surprising. As many a wise mentor has said, the way to build up our agency is to do hard things - and who enjoys that? Responding to the invitation to “life as a vocation” will not always be easy, and will often be frightening, requiring faith, hope, and love. It will be best sustained in community - real community with real humans, I hasten to add: elders who have developed wisdom by going ahead of us, peers with whom to empathize about the challenges joys of the adventure of being human. For Christians, we rely on Christ’s promise to be present with us, to the end of the age, and especially where we are gathered together in His name, as His people.
This invitation to a called life, rather than a self-planned life project, is not an optional add-on for exceptional humans; it is the core of what it means to be human. We may achieve greater efficiency by delegating our agency to artificial intelligence, but it will come at the cost of our humanity.
